
import pandas as pd
import matplotlib.pyplot as plt
import numpy as np
# Read dataset
bank_df= pd.read_csv("bankruptcy_prediction.csv")
# Remove any rows that are not finite or contain Nan values
bank_df = bank_df[np.isfinite(bank_df).all(1)]
# Shows the columns that have any Nan values
columns_with_na = bank_df.columns[bank_df.isna().any()].tolist()
print(columns_with_na)

[]


# Explanatory Variables
X = bank_df.iloc[:, 1:]
# Response Variable
y = bank_df.iloc[:, 0]


X.describe()


# Change dtype of each column
bank_df['Bankrupt?'] = bank_df['Bankrupt?'].astype(float)


num_net_income_flag = X.groupby(X[" Net Income Flag"]).size()
num_liability_assets_flag = X.groupby(X[" Liability-Assets Flag"]).size()
print(num_net_income_flag)
print(num_liability_assets_flag)

 Net Income Flag
1    6819
dtype: int64
 Liability-Assets Flag
0    6811
1       8
dtype: int64


X = X.drop(' Net Income Flag', axis=1)
X = X.drop(' Liability-Assets Flag', axis=1)
len(X.columns)

93


import seaborn as sns
# Plot correlation matrix using a heatmap
plt.figure(figsize=(30,30))
sns.heatmap(X.corr(), cmap='coolwarm')
plt.show()


from sklearn.model_selection import train_test_split
from sklearn.preprocessing import StandardScaler
# Function of StandardScaler
scaler = StandardScaler()
# Fit X and normalize the features
column_names = X.columns.tolist()
# Transform data using a scaler
scaler.fit(X)
X = scaler.transform(X)
# Create a data frame with the standardized data and column names
X = pd.DataFrame(X, columns=column_names)
# Split the data set into training and testing
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3, random_state=42, shuffle=True)


from sklearn.decomposition import PCA
"""pca_visualization function will reduce a dataset with n features to 3 components 
   and graph the components in three-dimensional space while classifying each data 
   point in color by its response variable."""
def pca_visualization(X, y, elevation):
    # Colors for the graph
    light_green = '#90EE90' 
    dark_blue = '#00008B' 
    # PCA declaration
    pca = PCA(n_components=3)
    X3D = pca.fit_transform(X)
    df_pca = pd.DataFrame(X3D, columns=['PC1', 'PC2', 'PC3'])
    # Add the y to the dataset
    y = y.to_frame('Bankruptcy').reset_index(drop=True)
    df_pca = pd.concat([df_pca, y], axis=1)
    # Plot the PCA components, highlighting them based on the classification of the response variable
    fig = plt.figure(figsize=(20, 20))
    ax = fig.add_subplot(projection='3d')
    # Graph the components
    ax.scatter(df_pca[df_pca['Bankruptcy'] == 0]['PC1'], df_pca[df_pca['Bankruptcy'] == 0]['PC2'], 
               df_pca[df_pca['Bankruptcy'] == 0]['PC3'], c = light_green, label = 'Non-Bankrupt')
    # Overlap the Bankrupt with Non-Bankrupt instances
    ax.scatter(df_pca[df_pca['Bankruptcy'] == 1]['PC1'], df_pca[df_pca['Bankruptcy'] == 1]['PC2'], 
               df_pca[df_pca['Bankruptcy'] == 1]['PC3'], c = dark_blue, label = 'Bankrupt')
    ax.view_init(elev=elevation, azim=50)
    ax.set_xlabel('PC1')
    ax.set_ylabel('PC2')
    ax.set_zlabel('PC3')
    ax.legend()
    plt.show()


pca_visualization(X, y, 70)


pca = PCA()
pca.fit(bank_df)
cumsum = np.cumsum(pca.explained_variance_ratio_)
# Capturing at least 95% of the variability in the dataset
d = np.argmax(cumsum >= 0.95) + 1
print("n Dimensions: " + str(d))

n Dimensions: 8


columns_ = bank_df.columns
num_pc = d
# Iterate through each PC to find the largest value that accounts for variance
for value in np.arange(num_pc):
    input_list = pca.components_[value]
    index = [index for index, item in enumerate(input_list) if item == input_list.max()]
    print("PC" + str(value + 1) + " " + str(columns_[index][0] + ": " + str(input_list[index][0])))

PC1  Quick Asset Turnover Rate: 0.6713563367601042
PC2  Inventory Turnover Rate (times): 0.7698179512194367
PC3  Operating Expense Rate: 0.812649178810887
PC4  Total Asset Growth Rate: 0.6407899938064151
PC5  Cash Turnover Rate: 0.6320250187572218
PC6  Research and development expense rate: 0.904031963105138
PC7  Fixed Assets Turnover Frequency: 0.9587583705789046
PC8  Current Asset Turnover Rate: 0.8216726046347916


from mpl_toolkits.mplot3d import Axes3D
from sklearn.metrics import pairwise_distances
from sklearn.manifold import MDS
# Define the label 
labels = list(X.columns)
# Normal the corr values
corrmat = X.corr()
corr_dmatrix = 1 - corrmat
# Create MDS to 3D
model_bank = MDS(n_components=3, dissimilarity='precomputed', random_state=1)
mds_out_bank = model_bank.fit_transform(corr_dmatrix)
# Reference PCA 1 and 2 for the appropriate axis
X3D = mds_out_bank[:,0]
Y3D = mds_out_bank[:,1]
Z3D = mds_out_bank[:,2]
# Create the graph 
fig = plt.figure(figsize=(25, 25))
ax = fig.add_subplot(projection='3d')
ax.set_aspect('auto')
ax.view_init(elev=20, azim=50)
ax.scatter3D(X3D, Y3D, Z3D)
# Labels for data points
for i, label in enumerate(labels):
    ax.text(X3D[i], Y3D[i], Z3D[i], label, fontsize=10)


"""count_bankruptcies function will graphically represent the class imbalance
   between Bankrupt and Non-Bankrupt instances."""
def count_bankruptcies(y_): 
    # Count the number of Bankrupt and Non-Bankrupt instances 
    num_bankruptcies = y_.groupby(y_).count()
    plot = num_bankruptcies.plot(kind="bar", color="darkgreen")
    # Set the labels
    plot.set_xticklabels(['Non-Bankrupt', 'Bankrupt'], rotation=0)
    plt.ylabel('Count')
    # % of Bankrupt instances
    print(num_bankruptcies)
    print("\n% of Bankruptcy Occurrences: {:.0%}".format(num_bankruptcies[1] / (num_bankruptcies[0] + num_bankruptcies[1])))


# Function that will create plot of Bankruptcies vs Non-Bankruptcies
count_bankruptcies(y)

Bankrupt?
0    6599
1     220
Name: Bankrupt?, dtype: int64

% of Bankruptcy Occurrences: 3%


# Function that will create a plot of Bankruptcies vs. Non-Bankruptcies
count_bankruptcies(y_train)

Bankrupt?
0    4631
1     142
Name: Bankrupt?, dtype: int64

% of Bankruptcy Occurrences: 3%


from imblearn.over_sampling import ADASYN
print("Current Number of Training Observations: {}".format(len(X_train)))

Current Number of Training Observations: 4773


adasyn = ADASYN(random_state=42)
# Perform oversampling on the training set
X_train_resampled_ADASYN, y_train_resampled_ADASYN = adasyn.fit_resample(X_train, y_train)


print("New Number of Training Observations: {}".format(len(X_train_resampled_ADASYN)))

New Number of Training Observations: 9309


# Function that will create a plot of Bankruptcies vs. Non-Bankruptcies
count_bankruptcies(y_train_resampled_ADASYN)

Bankrupt?
0    4631
1    4678
Name: Bankrupt?, dtype: int64

% of Bankruptcy Occurrences: 50%


pca_visualization(X_train_resampled_ADASYN, y_train_resampled_ADASYN, 50)


df = pd.concat([X_train, y_train], axis=1)


# Separate minority and majority class samples
minority_class = df[df["Bankrupt?"] == 1]  
majority_class = df[df["Bankrupt?"] == 0] 
# Oversample the minority class
oversampled_minority_class = minority_class.sample(n=len(majority_class), replace=True, random_state=42)
# Concatenate the oversampled minority class with the majority class
df_oversampled = pd.concat([oversampled_minority_class, majority_class], axis=0)
# X and y split
y_n_oversampling = df_oversampled["Bankrupt?"]
X_n_oversampling = df_oversampled.drop("Bankrupt?", axis=1)


# Function that will create plot of Bankruptcies vs Non-Bankruptcies
count_bankruptcies(y_n_oversampling)

Bankrupt?
0    4631
1    4631
Name: Bankrupt?, dtype: int64

% of Bankruptcy Occurrences: 50%


# Function that will create plot of Bankruptcies vs Non-Bankruptcies
pca_visualization(X_n_oversampling, y_n_oversampling, 40)


from imblearn.under_sampling import RandomUnderSampler
# Calling the undersampling function
rus = RandomUnderSampler(random_state=24)
X_resampled_undersampled, y_resampled_undersampled = rus.fit_resample(X_train, y_train)


# Function that will create a plot of Bankruptcies vs. Non-Bankruptcies
count_bankruptcies(y_resampled_undersampled)

Bankrupt?
0    142
1    142
Name: Bankrupt?, dtype: int64

% of Bankruptcy Occurrences: 50%


pca_visualization(X_resampled_undersampled, y_resampled_undersampled, 40)


from imblearn.under_sampling import NearMiss
# NearMiss-1
nm1 = NearMiss(version=1)
X_resampled_nm1, y_resampled_nm1 = nm1.fit_resample(X_train, y_train)


count_bankruptcies(y_resampled_nm1)

Bankrupt?
0    142
1    142
Name: Bankrupt?, dtype: int64

% of Bankruptcy Occurrences: 50%


pca_visualization(X_resampled_nm1, y_resampled_nm1, 40)


# NearMiss-2
nm2 = NearMiss(version=2)
X_resampled_nm2, y_resampled_nm2 = nm2.fit_resample(X_train, y_train)


count_bankruptcies(y_resampled_nm2)

Bankrupt?
0    142
1    142
Name: Bankrupt?, dtype: int64

% of Bankruptcy Occurrences: 50%


pca_visualization(X_resampled_nm2, y_resampled_nm2, 40)


# NearMiss-3
nm3 = NearMiss(version=3)
X_resampled_nm3, y_resampled_nm3 = nm3.fit_resample(X_train, y_train)


# Function that will create a plot of Bankruptcies vs. Non-Bankruptcies
count_bankruptcies(y_resampled_nm3)

Bankrupt?
0    142
1    142
Name: Bankrupt?, dtype: int64

% of Bankruptcy Occurrences: 50%


pca_visualization(X_resampled_nm3, y_resampled_nm3, 40)


from sklearn.neighbors import KNeighborsClassifier
from sklearn.svm import SVC
from sklearn.linear_model import LogisticRegression
from xgboost import XGBClassifier
from sklearn.metrics import accuracy_score, roc_auc_score, confusion_matrix, roc_curve, auc
from sklearn.model_selection import GridSearchCV
from sklearn.linear_model import RidgeClassifier


# List of models
models = [KNeighborsClassifier(), RidgeClassifier(random_state=2), SVC(random_state=7), XGBClassifier(random_state=4)]
# List of X and y
inputs = [[X_train, y_train], [X_train_resampled_ADASYN, y_train_resampled_ADASYN], 
          [X_n_oversampling, y_n_oversampling], [X_resampled_undersampled, y_resampled_undersampled], 
          [X_resampled_nm3, y_resampled_nm3], [X_resampled_nm2, y_resampled_nm2], [X_resampled_nm1, y_resampled_nm1]]


# KNN Parameters
knn_params = {
    'n_neighbors': [3, 5, 7], 
    'weights': ['uniform', 'distance'], 
    'metric': ['euclidean', 'manhattan', 'minkowski']
}
# Ridge Classification Parameters
rc_params = {
    'alpha': [0.01, 0.1, 10.0],
} 
# SVC Parameters
svc_params = {
    'C': [0.1, 1, 10] 
}
# XBG
xgb_params = {
    'n_estimators': [50, 100, 200], 
    'max_depth': [3, 5, 7], 
    'learning_rate': [0.01, 0.1, 0.2]
}


parameters = [knn_params, rc_params, svc_params, xgb_params]


import warnings
warnings.filterwarnings("ignore")


# Dataframe to save relevant information from trained models
df_outcome = pd.DataFrame({'Model': [], 'Parameters': [], 
                   'Scaling Type': [], 'Accuracy': [], 'Specificity': [], 
                   'Sensitivity': [], 'AUC Score': [], 'AUC Coords': []})


# Calculate the specificity and sensitivity score
def specificity_score(y_test, y_pred):
    tp, fn, fp, tn = confusion_matrix(y_test, y_pred).ravel()
    return tn / (tn + fp)
# Calculate the specificity and sensitivity score
def sensitivity_score(y_test, y_pred):
    tp, fn, fp, tn = confusion_matrix(y_test, y_pred).ravel()
    return tp / (tp + fn)


import itertools
"""Will iterate through each model and data scaling method to find the performance based on three metrics for all model variations."""
for model_type, tuning in zip(models, parameters):
    # It will grab the name of the model
    model = str(model_type)
    count = len(list(itertools.takewhile(lambda char: char != "(", model)))    
    model_name = model[:count]
    # Iterate through every type of sample selection
    for index, input_type in enumerate(inputs):
        # Grid search to tune the model
        grid_search = GridSearchCV(model_type, tuning, cv=2)
        grid_search.fit(input_type[0], input_type[1])
        y_pred = grid_search.predict(X_test) 
        # Classification Metrics
        accuracy = accuracy_score(y_test, y_pred)
        specificity = specificity_score(y_test, y_pred)
        sensitivity = sensitivity_score(y_test, y_pred)
        # Conditional statement that determines the best sample type per model and tuning parameters
        fpr, tpr, _ = roc_curve(y_test, y_pred, drop_intermediate=False)
        auc_coords = [fpr, tpr]
        auc_score = auc(fpr, tpr)
        # The parameters of the best grid search
        parameters = str(grid_search.best_estimator_)
        sample_type = index
        # New row will be added to data frame 
        new_row = {'Model': model_name, 'Parameters': parameters, 
                   'Scaling Type': index, 'Accuracy': accuracy, 'Specificity': specificity, 
                   'Sensitivity': sensitivity, 'AUC Score': auc_score, 'AUC Coords': auc_coords}
        df_outcome = df_outcome.append(new_row, ignore_index=True)


# Map to change the Sample types to the appropriate name
dict_map = {
    0.0: "Normal", 
    1.0: "ADASYN", 
    2.0: "Naïve Oversampling", 
    3.0: "Random Undersampling",
    4.0: "NearMiss-3", 
    5.0: "NearMiss-2",
    6.0: "NearMiss-1"
}
# Change display to show all information
pd.options.display.max_colwidth = None 
pd.options.display.expand_frame_repr = False
# Change the index to Model column
df_outcome["Scaling Type"] = df_outcome["Scaling Type"].map(dict_map)
ranked_df_outcome = df_outcome.sort_values(by='AUC Score', ascending=False)


# Print the top 3 models based on AUC
ranked_df_outcome = ranked_df_outcome.reset_index().drop("index", axis=1)
ranked_df_outcome


plt.figure()
# PLot the top 3 highest AUC curves
plt.plot(ranked_df_outcome["AUC Coords"][0][0], ranked_df_outcome["AUC Coords"][0][1], 
         color='green', lw=2, label='SVC: ROC curve (area = %0.2f)' % ranked_df_outcome["AUC Score"][0])
plt.plot(ranked_df_outcome["AUC Coords"][1][0], ranked_df_outcome["AUC Coords"][1][1], 
         color='blue', lw=2, label='Ridge Classifier: ROC curve (area = %0.2f)' % ranked_df_outcome["AUC Score"][1])
plt.plot(ranked_df_outcome["AUC Coords"][2][0], ranked_df_outcome["AUC Coords"][2][1], 
         color='red', lw=2, label='Ridge Classifier: ROC curve (area = %0.2f)' % ranked_df_outcome["AUC Score"][2])
# Graph will include the x and y axis
plt.plot([0, 1], [0, 1], color='black', lw=2, linestyle='--')
plt.xlim([0.0, 1.0])
plt.ylim([0.0, 1.05])
plt.xlabel('False Positive Rate')
plt.ylabel('True Positive Rate')
plt.title('ROC Graph')
plt.legend(loc="lower right")
plt.show()


# Fit RC based on its best tuned parameters
full_model = RidgeClassifier(alpha=0.1, random_state=2)	
full_model.fit(X_n_oversampling, y_n_oversampling)
y_pred = full_model.predict(X_test)
np.set_printoptions(suppress=True)
print("Intercept:", full_model.intercept_)
print("Coefficients:", full_model.coef_[0])

Intercept: [-0.52010447]
Coefficients: [ 0.20228856  0.09440332 -0.34723735 -0.76883745  1.39671364  0.15714751
  0.20186476  0.17074391  0.01143775 -0.37647543  0.03678587  0.00719175
 -0.09029039  0.00846217 -0.04424142 -1.39459704  0.51387286  0.85610312
 -0.15516041  0.07593933  0.04664507  0.36646357  0.02168419  0.00174802
  0.00079315 -0.08019903  0.13189996 -0.02014327 -0.02902959 -0.01364309
 -0.00211813  0.10658514 -0.00000055  0.00649963 -0.00756427  0.02851717
  0.09499013 -0.09499013  0.00893141  0.07131265  0.03117589 -0.17066479
 -0.12070152 -0.38266254  0.00065388 -0.04371326 -0.02564506 -0.01175764
  0.0923962  -0.16185248  0.00701409 -0.02912565 -0.01823177 -0.06914183
 -0.03445078  0.04491898 -0.05524929 -0.05068682  0.01351779  0.15082919
  0.1568896  -0.02387505  0.01041355  0.01674454  0.38074026  0.33426722
  0.00542598  0.02914988 -0.26804209 -0.03376001 -0.03736823  0.00243052
 -0.14390421 -0.14219752  0.09080621  0.00000279  0.01674454  0.33426722
  0.13741894  0.01975731 -0.07158448 -0.22607418 -0.09544591  0.05002132
 -0.08955298 -0.02227031  0.00897177 -0.62512708  0.0349217  -0.30008228
  0.02699974 -0.00862049 -0.02156868]


import seaborn as sns
# Print confusion matrix for optimal model
x_labels = ['Non-Bankrupt', 'Bankrupt'] 
y_labels = ['Non-Bankrupt', 'Bankrupt']
sns.heatmap(confusion_matrix(y_test, y_pred), annot=True, cbar=False, fmt="d", 
            xticklabels=x_labels, yticklabels=y_labels)
# Set axis labels
plt.xlabel('Predicted')
plt.ylabel('Actual')
plt.title('Confusion Matrix')
plt.show()


print("Specificity: {:.0%}".format(66/(12+66)))  
print("Sensitivity: {:.0%}".format(1697/(1697+271)))  
print("Accuracy: {:.0%}".format(accuracy_score(y_test, y_pred)))

Specificity: 85%
Sensitivity: 86%
Accuracy: 86%


from sklearn.feature_selection import SequentialFeatureSelector
# Size of the reduced model
n_features_to_select = [10, 20, 30, 40, 50, 60, 70, 80, 90]
# Selected columns per n_features_to_select
selected_columns = []
# Get the selected feature names per n_feature_to_select
for n_features in n_features_to_select:
    # Declare forward selection function selection
    sfs = SequentialFeatureSelector(full_model, n_features_to_select=n_features, direction='forward')
    # Fit the model using all of the data available since we want to find the subset of 
    # Critical features for the prediction of bankruptcy
    sfs.fit(X, y)
    # Get the selected feature indices
    selected_feature_indices = sfs.get_support(indices=True)
    selected_column_names = list(X.columns[selected_feature_indices])
    selected_columns.append(selected_column_names)
    # Print the selected feature names
    print("{} Selected Features: {}\n".format(n_features, selected_column_names))

10 Selected Features: [' ROA(C) before interest and depreciation before interest', ' ROA(A) before interest and % after tax', ' Operating Gross Margin', ' Realized Sales Gross Margin', ' Operating Profit Rate', ' Operating Expense Rate', ' Research and development expense rate', ' Per Share Net profit before tax (Yuan ¥)', ' Net Value Growth Rate', ' Working Capital/Equity']

20 Selected Features: [' ROA(C) before interest and depreciation before interest', ' ROA(A) before interest and % after tax', ' Operating Gross Margin', ' Realized Sales Gross Margin', ' Operating Profit Rate', ' Operating Expense Rate', ' Research and development expense rate', ' Cash flow rate', ' Interest-bearing debt interest rate', ' Tax rate (A)', ' Net Value Per Share (B)', ' Net Value Per Share (A)', ' Net Value Per Share (C)', ' Cash Flow Per Share', ' Revenue Per Share (Yuan ¥)', ' Operating Profit Per Share (Yuan ¥)', ' Per Share Net profit before tax (Yuan ¥)', ' After-tax Net Profit Growth Rate', ' Net Value Growth Rate', ' Working Capital/Equity']

30 Selected Features: [' ROA(C) before interest and depreciation before interest', ' ROA(A) before interest and % after tax', ' Operating Gross Margin', ' Realized Sales Gross Margin', ' Operating Profit Rate', ' Operating Expense Rate', ' Research and development expense rate', ' Cash flow rate', ' Interest-bearing debt interest rate', ' Tax rate (A)', ' Net Value Per Share (B)', ' Net Value Per Share (A)', ' Net Value Per Share (C)', ' Cash Flow Per Share', ' Revenue Per Share (Yuan ¥)', ' Operating Profit Per Share (Yuan ¥)', ' Per Share Net profit before tax (Yuan ¥)', ' After-tax Net Profit Growth Rate', ' Regular Net Profit Growth Rate', ' Continuous Net Profit Growth Rate', ' Total Asset Growth Rate', ' Net Value Growth Rate', ' Total Asset Return Growth Rate Ratio', ' Cash Reinvestment %', ' Current Ratio', ' Interest Expense Ratio', ' Total debt/Total net worth', ' Long-term fund suitability ratio (A)', ' Net profit before tax/Paid-in capital', ' Working Capital/Equity']

40 Selected Features: [' ROA(C) before interest and depreciation before interest', ' ROA(A) before interest and % after tax', ' Operating Gross Margin', ' Realized Sales Gross Margin', ' Operating Profit Rate', ' Operating Expense Rate', ' Research and development expense rate', ' Cash flow rate', ' Interest-bearing debt interest rate', ' Tax rate (A)', ' Net Value Per Share (B)', ' Net Value Per Share (A)', ' Net Value Per Share (C)', ' Cash Flow Per Share', ' Revenue Per Share (Yuan ¥)', ' Operating Profit Per Share (Yuan ¥)', ' Per Share Net profit before tax (Yuan ¥)', ' After-tax Net Profit Growth Rate', ' Regular Net Profit Growth Rate', ' Continuous Net Profit Growth Rate', ' Total Asset Growth Rate', ' Net Value Growth Rate', ' Total Asset Return Growth Rate Ratio', ' Cash Reinvestment %', ' Current Ratio', ' Interest Expense Ratio', ' Total debt/Total net worth', ' Long-term fund suitability ratio (A)', ' Operating profit/Paid-in capital', ' Net profit before tax/Paid-in capital', ' Total Asset Turnover', ' Accounts Receivable Turnover', ' Average Collection Days', ' Inventory Turnover Rate (times)', ' Fixed Assets Turnover Frequency', ' Operating profit per person', ' Allocation rate per person', ' Current Assets/Total Assets', ' Cash/Total Assets', ' Working Capital/Equity']

50 Selected Features: [' ROA(C) before interest and depreciation before interest', ' ROA(A) before interest and % after tax', ' ROA(B) before interest and depreciation after tax', ' Operating Gross Margin', ' Realized Sales Gross Margin', ' Operating Profit Rate', ' Operating Expense Rate', ' Research and development expense rate', ' Cash flow rate', ' Interest-bearing debt interest rate', ' Tax rate (A)', ' Net Value Per Share (B)', ' Net Value Per Share (A)', ' Net Value Per Share (C)', ' Cash Flow Per Share', ' Revenue Per Share (Yuan ¥)', ' Operating Profit Per Share (Yuan ¥)', ' Per Share Net profit before tax (Yuan ¥)', ' After-tax Net Profit Growth Rate', ' Regular Net Profit Growth Rate', ' Continuous Net Profit Growth Rate', ' Total Asset Growth Rate', ' Net Value Growth Rate', ' Total Asset Return Growth Rate Ratio', ' Cash Reinvestment %', ' Current Ratio', ' Interest Expense Ratio', ' Total debt/Total net worth', ' Long-term fund suitability ratio (A)', ' Operating profit/Paid-in capital', ' Net profit before tax/Paid-in capital', ' Total Asset Turnover', ' Accounts Receivable Turnover', ' Average Collection Days', ' Inventory Turnover Rate (times)', ' Fixed Assets Turnover Frequency', ' Operating profit per person', ' Allocation rate per person', ' Working Capital to Total Assets', ' Current Assets/Total Assets', ' Cash/Total Assets', ' Quick Assets/Current Liability', ' Operating Funds to Liability', ' Inventory/Working Capital', ' Inventory/Current Liability', ' Working Capital/Equity', ' Long-term Liability to Current Assets', ' Retained Earnings to Total Assets', ' Cash Flow to Total Assets', ' Net Income to Total Assets']

60 Selected Features: [' ROA(C) before interest and depreciation before interest', ' ROA(A) before interest and % after tax', ' ROA(B) before interest and depreciation after tax', ' Operating Gross Margin', ' Realized Sales Gross Margin', ' Operating Profit Rate', ' Operating Expense Rate', ' Research and development expense rate', ' Cash flow rate', ' Interest-bearing debt interest rate', ' Tax rate (A)', ' Net Value Per Share (B)', ' Net Value Per Share (A)', ' Net Value Per Share (C)', ' Cash Flow Per Share', ' Revenue Per Share (Yuan ¥)', ' Operating Profit Per Share (Yuan ¥)', ' Per Share Net profit before tax (Yuan ¥)', ' After-tax Net Profit Growth Rate', ' Regular Net Profit Growth Rate', ' Continuous Net Profit Growth Rate', ' Total Asset Growth Rate', ' Net Value Growth Rate', ' Total Asset Return Growth Rate Ratio', ' Cash Reinvestment %', ' Current Ratio', ' Interest Expense Ratio', ' Total debt/Total net worth', ' Long-term fund suitability ratio (A)', ' Operating profit/Paid-in capital', ' Net profit before tax/Paid-in capital', ' Total Asset Turnover', ' Accounts Receivable Turnover', ' Average Collection Days', ' Inventory Turnover Rate (times)', ' Fixed Assets Turnover Frequency', ' Net Worth Turnover Rate (times)', ' Operating profit per person', ' Allocation rate per person', ' Working Capital to Total Assets', ' Current Assets/Total Assets', ' Cash/Total Assets', ' Quick Assets/Current Liability', ' Current Liability to Assets', ' Operating Funds to Liability', ' Inventory/Working Capital', ' Inventory/Current Liability', ' Working Capital/Equity', ' Long-term Liability to Current Assets', ' Retained Earnings to Total Assets', ' Current Asset Turnover Rate', ' Quick Asset Turnover Rate', ' Cash Turnover Rate', ' Fixed Assets to Assets', ' Cash Flow to Total Assets', ' Cash Flow to Liability', ' Net Income to Total Assets', ' Total assets to GNP price', ' No-credit Interval', ' Gross Profit to Sales']

70 Selected Features: [' ROA(C) before interest and depreciation before interest', ' ROA(A) before interest and % after tax', ' ROA(B) before interest and depreciation after tax', ' Operating Gross Margin', ' Realized Sales Gross Margin', ' Operating Profit Rate', ' Pre-tax net Interest Rate', ' After-tax net Interest Rate', ' Non-industry income and expenditure/revenue', ' Continuous interest rate (after tax)', ' Operating Expense Rate', ' Research and development expense rate', ' Cash flow rate', ' Interest-bearing debt interest rate', ' Tax rate (A)', ' Net Value Per Share (B)', ' Net Value Per Share (A)', ' Net Value Per Share (C)', ' Persistent EPS in the Last Four Seasons', ' Cash Flow Per Share', ' Revenue Per Share (Yuan ¥)', ' Operating Profit Per Share (Yuan ¥)', ' Per Share Net profit before tax (Yuan ¥)', ' Operating Profit Growth Rate', ' After-tax Net Profit Growth Rate', ' Regular Net Profit Growth Rate', ' Continuous Net Profit Growth Rate', ' Total Asset Growth Rate', ' Net Value Growth Rate', ' Total Asset Return Growth Rate Ratio', ' Cash Reinvestment %', ' Current Ratio', ' Interest Expense Ratio', ' Total debt/Total net worth', ' Long-term fund suitability ratio (A)', ' Operating profit/Paid-in capital', ' Net profit before tax/Paid-in capital', ' Total Asset Turnover', ' Accounts Receivable Turnover', ' Average Collection Days', ' Inventory Turnover Rate (times)', ' Fixed Assets Turnover Frequency', ' Net Worth Turnover Rate (times)', ' Operating profit per person', ' Allocation rate per person', ' Working Capital to Total Assets', ' Current Assets/Total Assets', ' Cash/Total Assets', ' Quick Assets/Current Liability', ' Current Liability to Assets', ' Operating Funds to Liability', ' Inventory/Working Capital', ' Inventory/Current Liability', ' Working Capital/Equity', ' Long-term Liability to Current Assets', ' Retained Earnings to Total Assets', ' Current Asset Turnover Rate', ' Quick Asset Turnover Rate', ' Cash Turnover Rate', ' Fixed Assets to Assets', ' Cash Flow to Total Assets', ' Cash Flow to Liability', ' Net Income to Total Assets', ' Total assets to GNP price', ' No-credit Interval', ' Gross Profit to Sales', ' Liability to Equity', ' Degree of Financial Leverage (DFL)', ' Interest Coverage Ratio (Interest expense to EBIT)', ' Equity to Liability']

80 Selected Features: [' ROA(C) before interest and depreciation before interest', ' ROA(A) before interest and % after tax', ' ROA(B) before interest and depreciation after tax', ' Operating Gross Margin', ' Realized Sales Gross Margin', ' Operating Profit Rate', ' Pre-tax net Interest Rate', ' After-tax net Interest Rate', ' Non-industry income and expenditure/revenue', ' Continuous interest rate (after tax)', ' Operating Expense Rate', ' Research and development expense rate', ' Cash flow rate', ' Interest-bearing debt interest rate', ' Tax rate (A)', ' Net Value Per Share (B)', ' Net Value Per Share (A)', ' Net Value Per Share (C)', ' Persistent EPS in the Last Four Seasons', ' Cash Flow Per Share', ' Revenue Per Share (Yuan ¥)', ' Operating Profit Per Share (Yuan ¥)', ' Per Share Net profit before tax (Yuan ¥)', ' Realized Sales Gross Profit Growth Rate', ' Operating Profit Growth Rate', ' After-tax Net Profit Growth Rate', ' Regular Net Profit Growth Rate', ' Continuous Net Profit Growth Rate', ' Total Asset Growth Rate', ' Net Value Growth Rate', ' Total Asset Return Growth Rate Ratio', ' Cash Reinvestment %', ' Current Ratio', ' Interest Expense Ratio', ' Total debt/Total net worth', ' Long-term fund suitability ratio (A)', ' Borrowing dependency', ' Contingent liabilities/Net worth', ' Operating profit/Paid-in capital', ' Net profit before tax/Paid-in capital', ' Total Asset Turnover', ' Accounts Receivable Turnover', ' Average Collection Days', ' Inventory Turnover Rate (times)', ' Fixed Assets Turnover Frequency', ' Net Worth Turnover Rate (times)', ' Revenue per person', ' Operating profit per person', ' Allocation rate per person', ' Working Capital to Total Assets', ' Current Assets/Total Assets', ' Cash/Total Assets', ' Quick Assets/Current Liability', ' Cash/Current Liability', ' Current Liability to Assets', ' Operating Funds to Liability', ' Inventory/Working Capital', ' Inventory/Current Liability', ' Working Capital/Equity', ' Long-term Liability to Current Assets', ' Retained Earnings to Total Assets', ' Total income/Total expense', ' Total expense/Assets', ' Current Asset Turnover Rate', ' Quick Asset Turnover Rate', ' Cash Turnover Rate', ' Cash Flow to Sales', ' Fixed Assets to Assets', ' Cash Flow to Total Assets', ' Cash Flow to Liability', ' CFO to Assets', ' Cash Flow to Equity', ' Net Income to Total Assets', ' Total assets to GNP price', ' No-credit Interval', ' Gross Profit to Sales', ' Liability to Equity', ' Degree of Financial Leverage (DFL)', ' Interest Coverage Ratio (Interest expense to EBIT)', ' Equity to Liability']

90 Selected Features: [' ROA(C) before interest and depreciation before interest', ' ROA(A) before interest and % after tax', ' ROA(B) before interest and depreciation after tax', ' Operating Gross Margin', ' Realized Sales Gross Margin', ' Operating Profit Rate', ' Pre-tax net Interest Rate', ' After-tax net Interest Rate', ' Non-industry income and expenditure/revenue', ' Continuous interest rate (after tax)', ' Operating Expense Rate', ' Research and development expense rate', ' Cash flow rate', ' Interest-bearing debt interest rate', ' Tax rate (A)', ' Net Value Per Share (B)', ' Net Value Per Share (A)', ' Net Value Per Share (C)', ' Persistent EPS in the Last Four Seasons', ' Cash Flow Per Share', ' Revenue Per Share (Yuan ¥)', ' Operating Profit Per Share (Yuan ¥)', ' Per Share Net profit before tax (Yuan ¥)', ' Realized Sales Gross Profit Growth Rate', ' Operating Profit Growth Rate', ' After-tax Net Profit Growth Rate', ' Regular Net Profit Growth Rate', ' Continuous Net Profit Growth Rate', ' Total Asset Growth Rate', ' Net Value Growth Rate', ' Total Asset Return Growth Rate Ratio', ' Cash Reinvestment %', ' Current Ratio', ' Quick Ratio', ' Interest Expense Ratio', ' Total debt/Total net worth', ' Debt ratio %', ' Long-term fund suitability ratio (A)', ' Borrowing dependency', ' Contingent liabilities/Net worth', ' Operating profit/Paid-in capital', ' Net profit before tax/Paid-in capital', ' Inventory and accounts receivable/Net value', ' Total Asset Turnover', ' Accounts Receivable Turnover', ' Average Collection Days', ' Inventory Turnover Rate (times)', ' Fixed Assets Turnover Frequency', ' Net Worth Turnover Rate (times)', ' Revenue per person', ' Operating profit per person', ' Allocation rate per person', ' Working Capital to Total Assets', ' Quick Assets/Total Assets', ' Current Assets/Total Assets', ' Cash/Total Assets', ' Quick Assets/Current Liability', ' Cash/Current Liability', ' Current Liability to Assets', ' Operating Funds to Liability', ' Inventory/Working Capital', ' Inventory/Current Liability', ' Working Capital/Equity', ' Current Liabilities/Equity', ' Long-term Liability to Current Assets', ' Retained Earnings to Total Assets', ' Total income/Total expense', ' Total expense/Assets', ' Current Asset Turnover Rate', ' Quick Asset Turnover Rate', ' Working capitcal Turnover Rate', ' Cash Turnover Rate', ' Cash Flow to Sales', ' Fixed Assets to Assets', ' Current Liability to Equity', ' Equity to Long-term Liability', ' Cash Flow to Total Assets', ' Cash Flow to Liability', ' CFO to Assets', ' Cash Flow to Equity', ' Current Liability to Current Assets', ' Net Income to Total Assets', ' Total assets to GNP price', ' No-credit Interval', ' Gross Profit to Sales', " Net Income to Stockholder's Equity", ' Liability to Equity', ' Degree of Financial Leverage (DFL)', ' Interest Coverage Ratio (Interest expense to EBIT)', ' Equity to Liability']


from scipy import stats
# Select the X_train with the feature columns
for features in selected_columns:
    # Retrain the model with tuned parameters using the reduction in features
    reduced_model = RidgeClassifier(alpha=0.1, random_state=2)	
    reduced_model.fit(X_n_oversampling[features], y_n_oversampling)
    # Calculating appropriate classification metrics
    y_pred = reduced_model.predict(X_test[features])
    accuracy = accuracy_score(y_test, y_pred)
    specificity = specificity_score(y_test, y_pred)
    sensitivity = sensitivity_score(y_test, y_pred)
    # Print the summary statistics
    print("No. of Features {}:".format(len(features)))
    print("Accuracy: {}".format(accuracy))
    print("Specificity: {}".format(specificity))
    print("Sensitivity: {}".format(sensitivity))
    print("\n")

No. of Features 10:
Accuracy: 0.8274682306940372
Specificity: 0.7307692307692307
Sensitivity: 0.8313008130081301


No. of Features 20:
Accuracy: 0.8025415444770283
Specificity: 0.8205128205128205
Sensitivity: 0.801829268292683


No. of Features 30:
Accuracy: 0.8079178885630498
Specificity: 0.7948717948717948
Sensitivity: 0.8084349593495935


No. of Features 40:
Accuracy: 0.8172043010752689
Specificity: 0.8589743589743589
Sensitivity: 0.8155487804878049


No. of Features 50:
Accuracy: 0.8528836754643206
Specificity: 0.8461538461538461
Sensitivity: 0.853150406504065


No. of Features 60:
Accuracy: 0.8558162267839687
Specificity: 0.8076923076923077
Sensitivity: 0.8577235772357723


No. of Features 70:
Accuracy: 0.8484848484848485
Specificity: 0.8076923076923077
Sensitivity: 0.8501016260162602


No. of Features 80:
Accuracy: 0.8504398826979472
Specificity: 0.8076923076923077
Sensitivity: 0.8521341463414634


No. of Features 90:
Accuracy: 0.8607038123167156
Specificity: 0.8461538461538461
Sensitivity: 0.8612804878048781

	ROA(C) before interest and depreciation before interest	ROA(A) before interest and % after tax	ROA(B) before interest and depreciation after tax	Operating Gross Margin	Realized Sales Gross Margin	Operating Profit Rate	Pre-tax net Interest Rate	After-tax net Interest Rate	Non-industry income and expenditure/revenue	Continuous interest rate (after tax)	...	Net Income to Total Assets	Total assets to GNP price	No-credit Interval	Gross Profit to Sales	Net Income to Stockholder's Equity	Liability to Equity	Degree of Financial Leverage (DFL)	Interest Coverage Ratio (Interest expense to EBIT)	Net Income Flag	Equity to Liability
count	6819.000000	6819.000000	6819.000000	6819.000000	6819.000000	6819.000000	6819.000000	6819.000000	6819.000000	6819.000000	...	6819.000000	6.819000e+03	6819.000000	6819.000000	6819.000000	6819.000000	6819.000000	6819.000000	6819.0	6819.000000
mean	0.505180	0.558625	0.553589	0.607948	0.607929	0.998755	0.797190	0.809084	0.303623	0.781381	...	0.807760	1.862942e+07	0.623915	0.607946	0.840402	0.280365	0.027541	0.565358	1.0	0.047578
std	0.060686	0.065620	0.061595	0.016934	0.016916	0.013010	0.012869	0.013601	0.011163	0.012679	...	0.040332	3.764501e+08	0.012290	0.016934	0.014523	0.014463	0.015668	0.013214	0.0	0.050014
min	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	...	0.000000	0.000000e+00	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	1.0	0.000000
25%	0.476527	0.535543	0.527277	0.600445	0.600434	0.998969	0.797386	0.809312	0.303466	0.781567	...	0.796750	9.036205e-04	0.623636	0.600443	0.840115	0.276944	0.026791	0.565158	1.0	0.024477
50%	0.502706	0.559802	0.552278	0.605997	0.605976	0.999022	0.797464	0.809375	0.303525	0.781635	...	0.810619	2.085213e-03	0.623879	0.605998	0.841179	0.278778	0.026808	0.565252	1.0	0.033798
75%	0.535563	0.589157	0.584105	0.613914	0.613842	0.999095	0.797579	0.809469	0.303585	0.781735	...	0.826455	5.269777e-03	0.624168	0.613913	0.842357	0.281449	0.026913	0.565725	1.0	0.052838
max	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000	...	1.000000	9.820000e+09	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000	1.0	1.000000

	Model	Parameters	Scaling Type	Accuracy	Specificity	Sensitivity	AUC Score	AUC Coords
0	SVC	SVC(C=1, random_state=7)	Random Undersampling	0.817693	0.897436	0.814533	0.855984	[[0.0, 0.18546747967479674, 1.0], [0.0, 0.8974358974358975, 1.0]]
1	RidgeClassifier	RidgeClassifier(alpha=0.1, random_state=2)	Naïve Oversampling	0.861681	0.846154	0.862297	0.854225	[[0.0, 0.13770325203252032, 1.0], [0.0, 0.8461538461538461, 1.0]]
2	RidgeClassifier	RidgeClassifier(alpha=10.0, random_state=2)	Random Undersampling	0.817204	0.871795	0.815041	0.843418	[[0.0, 0.18495934959349594, 1.0], [0.0, 0.8717948717948718, 1.0]]
3	RidgeClassifier	RidgeClassifier(alpha=0.1, random_state=2)	ADASYN	0.843597	0.833333	0.844004	0.838669	[[0.0, 0.1559959349593496, 1.0], [0.0, 0.8333333333333334, 1.0]]
4	XGBClassifier	XGBClassifier(base_score=None, booster=None, callbacks=None,\n colsample_bylevel=None, colsample_bynode=None,\n colsample_bytree=None, early_stopping_rounds=None,\n enable_categorical=False, eval_metric=None, feature_types=None,\n gamma=None, gpu_id=None, grow_policy=None, importance_type=None,\n interaction_constraints=None, learning_rate=0.2, max_bin=None,\n max_cat_threshold=None, max_cat_to_onehot=None,\n max_delta_step=None, max_depth=3, max_leaves=None,\n min_child_weight=None, missing=nan, monotone_constraints=None,\n n_estimators=50, n_jobs=None, num_parallel_tree=None,\n predictor=None, random_state=4, ...)	Random Undersampling	0.834311	0.820513	0.834858	0.827685	[[0.0, 0.16514227642276422, 1.0], [0.0, 0.8205128205128205, 1.0]]
5	KNeighborsClassifier	KNeighborsClassifier(metric='manhattan', n_neighbors=3)	Random Undersampling	0.817693	0.833333	0.817073	0.825203	[[0.0, 0.18292682926829268, 1.0], [0.0, 0.8333333333333334, 1.0]]
6	KNeighborsClassifier	KNeighborsClassifier(metric='manhattan', n_neighbors=3, weights='distance')	ADASYN	0.911046	0.641026	0.921748	0.781387	[[0.0, 0.0782520325203252, 1.0], [0.0, 0.6410256410256411, 1.0]]
7	XGBClassifier	XGBClassifier(base_score=None, booster=None, callbacks=None,\n colsample_bylevel=None, colsample_bynode=None,\n colsample_bytree=None, early_stopping_rounds=None,\n enable_categorical=False, eval_metric=None, feature_types=None,\n gamma=None, gpu_id=None, grow_policy=None, importance_type=None,\n interaction_constraints=None, learning_rate=0.2, max_bin=None,\n max_cat_threshold=None, max_cat_to_onehot=None,\n max_delta_step=None, max_depth=5, max_leaves=None,\n min_child_weight=None, missing=nan, monotone_constraints=None,\n n_estimators=200, n_jobs=None, num_parallel_tree=None,\n predictor=None, random_state=4, ...)	NearMiss-1	0.685239	0.794872	0.680894	0.737883	[[0.0, 0.31910569105691056, 1.0], [0.0, 0.7948717948717948, 1.0]]
8	RidgeClassifier	RidgeClassifier(alpha=10.0, random_state=2)	NearMiss-1	0.702346	0.769231	0.699695	0.734463	[[0.0, 0.3003048780487805, 1.0], [0.0, 0.7692307692307693, 1.0]]
9	KNeighborsClassifier	KNeighborsClassifier(metric='manhattan', n_neighbors=3)	NearMiss-1	0.890029	0.551282	0.903455	0.727369	[[0.0, 0.09654471544715447, 1.0], [0.0, 0.5512820512820513, 1.0]]
10	SVC	SVC(C=10, random_state=7)	Naïve Oversampling	0.921799	0.500000	0.938516	0.719258	[[0.0, 0.061483739837398375, 1.0], [0.0, 0.5, 1.0]]
11	XGBClassifier	XGBClassifier(base_score=None, booster=None, callbacks=None,\n colsample_bylevel=None, colsample_bynode=None,\n colsample_bytree=None, early_stopping_rounds=None,\n enable_categorical=False, eval_metric=None, feature_types=None,\n gamma=None, gpu_id=None, grow_policy=None, importance_type=None,\n interaction_constraints=None, learning_rate=0.1, max_bin=None,\n max_cat_threshold=None, max_cat_to_onehot=None,\n max_delta_step=None, max_depth=7, max_leaves=None,\n min_child_weight=None, missing=nan, monotone_constraints=None,\n n_estimators=50, n_jobs=None, num_parallel_tree=None,\n predictor=None, random_state=4, ...)	NearMiss-3	0.876833	0.538462	0.890244	0.714353	[[0.0, 0.10975609756097561, 1.0], [0.0, 0.5384615384615384, 1.0]]
12	XGBClassifier	XGBClassifier(base_score=None, booster=None, callbacks=None,\n colsample_bylevel=None, colsample_bynode=None,\n colsample_bytree=None, early_stopping_rounds=None,\n enable_categorical=False, eval_metric=None, feature_types=None,\n gamma=None, gpu_id=None, grow_policy=None, importance_type=None,\n interaction_constraints=None, learning_rate=0.1, max_bin=None,\n max_cat_threshold=None, max_cat_to_onehot=None,\n max_delta_step=None, max_depth=5, max_leaves=None,\n min_child_weight=None, missing=nan, monotone_constraints=None,\n n_estimators=200, n_jobs=None, num_parallel_tree=None,\n predictor=None, random_state=4, ...)	ADASYN	0.956500	0.371795	0.979675	0.675735	[[0.0, 0.02032520325203252, 1.0], [0.0, 0.3717948717948718, 1.0]]
13	SVC	SVC(C=10, random_state=7)	ADASYN	0.933040	0.384615	0.954776	0.669696	[[0.0, 0.04522357723577236, 1.0], [0.0, 0.38461538461538464, 1.0]]
14	KNeighborsClassifier	KNeighborsClassifier(metric='manhattan', n_neighbors=3)	Naïve Oversampling	0.940371	0.371795	0.962907	0.667351	[[0.0, 0.03709349593495935, 1.0], [0.0, 0.3717948717948718, 1.0]]
15	XGBClassifier	XGBClassifier(base_score=None, booster=None, callbacks=None,\n colsample_bylevel=None, colsample_bynode=None,\n colsample_bytree=None, early_stopping_rounds=None,\n enable_categorical=False, eval_metric=None, feature_types=None,\n gamma=None, gpu_id=None, grow_policy=None, importance_type=None,\n interaction_constraints=None, learning_rate=0.2, max_bin=None,\n max_cat_threshold=None, max_cat_to_onehot=None,\n max_delta_step=None, max_depth=5, max_leaves=None,\n min_child_weight=None, missing=nan, monotone_constraints=None,\n n_estimators=200, n_jobs=None, num_parallel_tree=None,\n predictor=None, random_state=4, ...)	Naïve Oversampling	0.960899	0.320513	0.986280	0.653397	[[0.0, 0.013719512195121951, 1.0], [0.0, 0.32051282051282054, 1.0]]
16	SVC	SVC(C=10, random_state=7)	NearMiss-1	0.355816	0.884615	0.334858	0.609737	[[0.0, 0.6651422764227642, 1.0], [0.0, 0.8846153846153846, 1.0]]
17	RidgeClassifier	RidgeClassifier(alpha=0.01, random_state=2)	NearMiss-3	0.625611	0.576923	0.627541	0.602232	[[0.0, 0.3724593495934959, 1.0], [0.0, 0.5769230769230769, 1.0]]
18	XGBClassifier	XGBClassifier(base_score=None, booster=None, callbacks=None,\n colsample_bylevel=None, colsample_bynode=None,\n colsample_bytree=None, early_stopping_rounds=None,\n enable_categorical=False, eval_metric=None, feature_types=None,\n gamma=None, gpu_id=None, grow_policy=None, importance_type=None,\n interaction_constraints=None, learning_rate=0.2, max_bin=None,\n max_cat_threshold=None, max_cat_to_onehot=None,\n max_delta_step=None, max_depth=5, max_leaves=None,\n min_child_weight=None, missing=nan, monotone_constraints=None,\n n_estimators=50, n_jobs=None, num_parallel_tree=None,\n predictor=None, random_state=4, ...)	Normal	0.964321	0.205128	0.994411	0.599769	[[0.0, 0.005589430894308943, 1.0], [0.0, 0.20512820512820512, 1.0]]
19	SVC	SVC(C=10, random_state=7)	NearMiss-3	0.663734	0.500000	0.670224	0.585112	[[0.0, 0.32977642276422764, 1.0], [0.0, 0.5, 1.0]]
20	KNeighborsClassifier	KNeighborsClassifier(metric='euclidean')	Normal	0.961388	0.115385	0.994919	0.555152	[[0.0, 0.00508130081300813, 1.0], [0.0, 0.11538461538461539, 1.0]]
21	RidgeClassifier	RidgeClassifier(alpha=0.01, random_state=2)	Normal	0.960899	0.051282	0.996951	0.524117	[[0.0, 0.003048780487804878, 1.0], [0.0, 0.05128205128205128, 1.0]]
22	SVC	SVC(C=0.1, random_state=7)	Normal	0.961877	0.000000	1.000000	0.500000	[[0.0, 1.0], [0.0, 1.0]]
23	KNeighborsClassifier	KNeighborsClassifier(metric='manhattan', n_neighbors=3)	NearMiss-3	0.406647	0.589744	0.399390	0.494567	[[0.0, 0.600609756097561, 1.0], [0.0, 0.5897435897435898, 1.0]]
24	RidgeClassifier	RidgeClassifier(alpha=10.0, random_state=2)	NearMiss-2	0.174487	0.628205	0.156504	0.392355	[[0.0, 0.8434959349593496, 1.0], [0.0, 0.6282051282051282, 1.0]]
25	XGBClassifier	XGBClassifier(base_score=None, booster=None, callbacks=None,\n colsample_bylevel=None, colsample_bynode=None,\n colsample_bytree=None, early_stopping_rounds=None,\n enable_categorical=False, eval_metric=None, feature_types=None,\n gamma=None, gpu_id=None, grow_policy=None, importance_type=None,\n interaction_constraints=None, learning_rate=0.1, max_bin=None,\n max_cat_threshold=None, max_cat_to_onehot=None,\n max_delta_step=None, max_depth=7, max_leaves=None,\n min_child_weight=None, missing=nan, monotone_constraints=None,\n n_estimators=50, n_jobs=None, num_parallel_tree=None,\n predictor=None, random_state=4, ...)	NearMiss-2	0.077713	0.730769	0.051829	0.391299	[[0.0, 0.948170731707317, 1.0], [0.0, 0.7307692307692307, 1.0]]
26	KNeighborsClassifier	KNeighborsClassifier(metric='euclidean', n_neighbors=3)	NearMiss-2	0.157380	0.589744	0.140244	0.364994	[[0.0, 0.8597560975609756, 1.0], [0.0, 0.5897435897435898, 1.0]]
27	SVC	SVC(C=10, random_state=7)	NearMiss-2	0.084555	0.666667	0.061484	0.364075	[[0.0, 0.9385162601626016, 1.0], [0.0, 0.6666666666666666, 1.0]]